From d30a9a177592ad0c68622a6e859eaa0b9b648390 Mon Sep 17 00:00:00 2001
From: mithandir <mithandir@gmail.com>
Date: Wed, 26 Jun 2024 09:33:54 +0200
Subject: [PATCH] Process the last 48h of posts to re-try summarization

---
 src/main/java/ch/climbd/newsfeed/controller/MlController.java | 2 +-
 .../java/ch/climbd/newsfeed/controller/MongoController.java   | 4 ++--
 2 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/src/main/java/ch/climbd/newsfeed/controller/MlController.java b/src/main/java/ch/climbd/newsfeed/controller/MlController.java
index e727b9b..6f4a87e 100644
--- a/src/main/java/ch/climbd/newsfeed/controller/MlController.java
+++ b/src/main/java/ch/climbd/newsfeed/controller/MlController.java
@@ -27,7 +27,7 @@ public class MlController {
 
     @PostConstruct
     public void fixQueueAfterRestart() {
-        var todaysNews = mongo.findAllPostedToday();
+        var todaysNews = mongo.findLast100PostsPostedInTheLast48h();
         todaysNews.stream()
                 .filter(news -> news.getSummary() == null || news.getSummary().isBlank())
                 .filter(news -> news.getContent() != null && !news.getContent().isBlank() && news.getContent().length() > 1000)
diff --git a/src/main/java/ch/climbd/newsfeed/controller/MongoController.java b/src/main/java/ch/climbd/newsfeed/controller/MongoController.java
index 35556c6..f280a04 100644
--- a/src/main/java/ch/climbd/newsfeed/controller/MongoController.java
+++ b/src/main/java/ch/climbd/newsfeed/controller/MongoController.java
@@ -113,9 +113,9 @@ public List<NewsEntry> findAllFilterdBySite(String host) {
         return template.find(query, NewsEntry.class);
     }
 
-    public List<NewsEntry> findAllPostedToday() {
+    public List<NewsEntry> findLast100PostsPostedInTheLast48h() {
         Query query = new Query();
-        query.addCriteria(Criteria.where("publishedAt").gte(ZonedDateTime.now().truncatedTo(ChronoUnit.DAYS).toInstant()));
+        query.addCriteria(Criteria.where("publishedAt").gte(ZonedDateTime.now().minusDays(2).toInstant()));
         query.with(Sort.by(Sort.Direction.DESC, "publishedAt"));
         query.limit(100);
         query.maxTimeMsec(1000);